Rampant Misexpression in a Mimulus (Monkeyflower) Introgression Line Caused by Hybrid Sterility, Not Regulatory Divergence

Abstract

Divergence in gene expression regulation is common between closely related species and may give rise to incompatibilities in their hybrid progeny. In this study, we investigated the relationship between regulatory evolution within species and reproductive isolation between species. We focused on a well-studied case of hybrid sterility between two closely related yellow monkeyflower species, Mimulus guttatus and Mimulus nasutus, that is caused by two epistatic loci, hybrid male sterility 1 (hms1) and hybrid male sterility 2 (hms2). We compared genome-wide transcript abundance across male and female reproductive tissues (i.e., stamens and carpels) from four genotypes: M. guttatus, M. nasutus, and sterile and fertile progeny from an advanced M. nasutus-M. guttatus introgression line carrying the hms1-hms2 incompatibility. We observed substantial variation in transcript abundance between M. guttatus and M. nasutus, including distinct but overlapping patterns of tissue-biased expression, providing evidence for regulatory divergence between these species. We also found rampant genome-wide misexpression, but only in the affected tissues (i.e., stamens) of sterile introgression hybrids carrying incompatible alleles at hms1 and hms2. Examining patterns of allele-specific expression in sterile and fertile introgression hybrids, we found evidence for interspecific divergence in cis- and trans-regulation, including compensatory cis-trans mutations likely to be driven by stabilizing selection. Nevertheless, species divergence in gene regulatory networks cannot explain the vast majority of the gene misexpression we observe in Mimulus introgression hybrids, which instead likely manifests as a downstream consequence of sterility itself.

Keywords: Mimulus; Dobzhansky–Muller incompatibility; allele-specific expression; hybrid male sterility; monkeyflower; regulatory divergence.

Fig. 1.

Crossing scheme to generate RSB introgression hybrids. First, Mimulus nasutus SF5 and Mimulus guttatus IM62 were cross pollinated, yielding an SF5–IM62 F₁ that was subsequently backcrossed to SF5, yielding a first-generation backcross (BC₁) population. A pollen sterile individual from the BC₁ population (circled in red) was selected and backcrossed to SF5, yielding a first-generation RSB introgression line (RSB₁). This selective backcrossing was repeated for six more generations, producing an RSB₇ population. Roughly, 50% of the RSB₇ siblings are pollen STE because they carry an heterozygous introgression on chromosome 6 (with an incompatible IM62 allele at hms1) in an SF5 genomic background (that is fixed for the incompatible SF5 allele at hms2), whereas the other 50% are pollen FER because they carry only SF5 alleles at hms1 (in the same genomic background that is fixed for the incompatible SF5 allele at hms2). Whole transcriptome sequencing was performed on RNA extracted from three biological replicates each of stamens and carpels (gray box, bottom left) from four genotypes, M. nasutus SF5, M. guttatus IM62, FER RSB₇ introgression hybrid, and STE RSB₇ introgression hybrid, for a total of 24 samples, representing eight tissue-genotype categories. To obtain sufficient RNA for sequence library preparation, we dissected four to eight floral buds each biological replicate. D = meiotic drive locus (Fishman and Willis 2005).

Fig. 2.

Genome-wide expression pattern across samples. Shown is a MDS plot comparing gene expression across the 24 RNAseq samples in our data set. Distance between points represents the leading (i.e., largest absolute) log 2 fold-change (FC) transcript abundance between pairs of samples mapped onto two-dimensional space: the y-axis shows dimension 1 (dim 1) and the x-axis shows dimension two (dim 2). The colors and shapes represent the different genotype-tissue categories. IM62, Mimulus guttatus IM62 parent; SF5, Mimulus nasutus SF5 parent; FER, fertile RSB₇ introgression hybrid; STE, sterile RSB₇ introgression hybrid; FERcp, FER carpel; FERst, FER stamen; IM62cp, IM62 carpel; IM62st, IM62 stamen; SF5cp, SF5 carpel; SF5st, SF5 stamen; STEcp, STE carpel; STEst, STE stamen.

Fig. 3.

Parental tissue-biased gene expression. Scatterplot shows relative transcript abundance (log 2 fold-change) between stamen and carpel tissues in the SF5 and IM62 parents. Of the 20,431 genes expressed in our data set, 1,812 (8.9%) and 5,700 (27.9%) were significantly (log 2 FC > 1.25, FDR ≤ 0.05) stamen biased (green points in top right quadrant) and carpel biased (green points in bottom left quadrant), respectively, in both parents. An additional 546 (2.7%) and 1,339 (6.6%) genes were stamen biased and carpel biased, respectively, in SF5 only (blue points), whereas 615 (3%) and 2,228 (10.9%) genes were stamen biased and carpel biased, respectively, in IM62 only (yellow points). A few genes exhibited opposing tissue-biased expression patterns in SF5 and IM62 (purple points): 28 (0.1%) were stamen biased in SF5 and carpel biased in IM62, whereas 50 (0.2%) were carpel biased in SF5 and stamen biased in IM62. The remaining 8,113 (39.7%) genes were evenly expressed between parental tissues (gray points). IM62, Mimulus guttatus IM62 parent; SF5, Mimulus nasutus SF5 parent.

Fig. 4.

Gene expression differences across genotypes and tissues. Venn diagrams show counts of genes with significantly altered transcript abundance (log 2 fold-change > 1.25, FDR ≤ 0.05) in carpels and stamens across three pairwise comparisons: 1) STE versus FER, 2) STE versus SF5, and 3) FER versus SF5. SF5, Mimulus nasutus SF5 parent; FER, fertile RSB₇ introgression hybrid; STE, sterile RSB₇ introgression hybrid.

Fig. 5.

Genome-wide pattern of expression in FER introgression hybrids. Points represent relative transcript abundance (log 2 fold-change) between FER and SF5 on the x-axis and between FER and IM62 on the y-axis across (A, C) carpels and (B, D) stamens for (A, B) heterozygous genes in the chromosome 11 introgression and (C, D) homozygous background genes. Points are colored by expression class (see supplementary table S2, Supplementary Material online, for description). IM62, Mimulus guttatus IM62 parent; SF5, Mimulus nasutus SF5 parent; FER, fertile RSB₇ introgression hybrid.

Fig. 6.

Genome-wide pattern of expression in STE introgression hybrids. Points represent relative transcript abundance (log 2 fold-change) between STE and SF5 on the x-axis and between STE and IM62 on the y-axis across (A, C, E) carpels and (B, D, F) stamens for heterozygous genes in the (A B) chromosome 6 and (C, D) chromosome 11 introgressions, and (E, F) homozygous background genes. Points are colored by expression class (see supplementary table S2, Supplementary Material online, for description). IM62, Mimulus guttatus IM62 parent; SF5, Mimulus nasutus SF5 parent; STE, sterile RSB₇ introgression hybrid.

Fig. 7.

Classification of parental cis- and trans-regulatory divergence across heterozygous genes in FER and STE introgression hybrids. Points represent relative transcript abundance (log 2 fold-change) between the parents, SF5 and IM62, on the x-axis (log 2[SF5/IM62] parents) and relative abundance of SF5- and IM62-specific transcripts in the hybrids on the y-axis (log 2[SF5/IM62] hybrid) across (A, B) FER and (C–F) STE tissues (i.e., carpels and stamens) for heterozygous genes in the (C, D) chromosome 6 and (A–D) chromosome 11 introgression regions. The diagonal line indicates relative parental expression (i.e., log 2[SF5/IM62] parents) and relative allelic expression (i.e., log 2[SF5/IM62] hybrid) are equal. Points that fall along the diagonal line represent genes for whom parental expression differences are entirely explained by cis-regulatory divergence. Points are colored by regulatory divergence category. IM62, Mimulus guttatus IM62 parent; SF5, Mimulus nasutus SF5 parent; FER, fertile RSB₇ introgression hybrid; STE, sterile RSB₇ introgression hybrid.